This page is about where Odyssey 616 came from, what it is, what it is not, and what it does badly as much as what it does well - in my own opinion, of course.
It all started with a concert. On 12 April 2025 I went to see Marvel Infinity Saga with friends. I liked the Marvel universe without ever having followed it seriously: I picked out the characters I liked, as things came out, with no order and no regularity.
The concert was a bit of a trigger. I like marathons, I had already run a few on other sagas, so why not do this one. Every film, every series, every one-shot released by Marvel Studios.
Like a lot of people who take this on, I first leaned on the Excel files shared by the community. They are excellent, very thorough, and they give you a real thread through a universe this large.
But as the marathon went on I ran into their limits: some released titles were missing, the viewing order didn't always match the one I wanted to follow, and I wanted something more personal. I went through several of them, each with its own order, its own inclusions, its own judgement calls.
I'm a developer and I often (always) have a side project on the go. In early 2026, I got to work.
It started as a simple app to track my own progress. The project quickly outgrew that: today it is a small ecosystem of four applications all sitting on the same database.
Marathon tracks viewing. Map catalogues the universe's real and fictional locations. Characters looks at the characters, where they appear, and who has played them over the years. The Hub ties it all together and works as a brand platform more than anything else.
The project was born in early 2026, just as AI tools crossed a fairly clear threshold. It let me build something I actually wanted at a speed that would have been far harder to reach before, and at the same time put these ways of working to the test with no client at stake, so I could judge for myself how these models are evolving and where they are worth using.
I tried several models - Claude, Copilot, Gemini, Codex - and several agent frameworks: BMAD, Superpowers, Osmani's Agent Skills. The goal was never to hand development over blindly, but to understand what these tools genuinely bring, where they stop, and under what conditions they become worthwhile. I'm a lead dev, and I'd say a real project beats an opinion formed from reading about it.
The hardest part was neither the interface nor the architecture - it was the data. The main source is TMDB. On paper you just call an API; in practice, the first pulls brought back wrong posters, irrelevant characters and badly matched titles. It took a lot of filtering, and a long stretch of correcting by hand.
The viewing order, the mapping of locations and the classification of universes are my own editorial work. The metadata - posters, synopses, dates - comes from TMDB, whose API makes this project possible.
The rest belongs to Marvel and Disney, who fiercely protect their universe, their characters and their artwork. This project was never meant to be an official product, a commercial platform, or a roundabout way of using proprietary artwork. It's a personal tool built around data and metadata, to explore a universe I enjoy.
Data kept by hand rather than scraped: locations are placed one by one, real and fictional alike, and the viewing order is settled title by title.
Fully translated into three languages, no ads, no tracking, no mandatory account. Tracking works on a phone, and offline.
Contradicting the first point of the previous section, the database isn't perfect. Errors remain, along with edge cases and entries still to enrich: it's reliable for what it serves, not exhaustive.
Series are tracked at the series level, not the episode. Some titles have no established chronological order and carry an estimate, flagged as such. Fictional locations are placed where they make sense, not where they are - there are no coordinates for Wakanda, for instance, even though we know it is on Earth.
And the interface makes no visual claims. I have no particular affinity for aesthetics and design, and that was never the point of the project: the screens are there to move through the data without friction. Yes, it looks like every other app churned out by AI. Never mind - that isn't the goal.
The marathon goes on, the data keeps changing, and the ecosystem still has plenty left to explore.
There is no real roadmap and no fixed plan: it moves along with whatever I feel like doing, and whatever I want to try out here and there. If you have ideas, feel free to write to me at [email protected] - always happy to discuss them.
The project runs on a domain name, hosting, and an API that's free but rate-limited. There is a Ko-fi for anyone who wants to chip in towards the costs - nothing mandatory, obviously.